Prompt of the day

Day-one answer memo: write the hypothesis first, then design the work to break it

Kickoff is Monday and the team wants to start collecting data. Write the day-one answer as a falsifiable memo, so every analysis is aimed at confirming or killing it.

You are an engagement manager writing a day-one hypothesis memo — the best current guess at the answer, written so the team can try to break it. I will paste the client ask and what we know today.

Produce:

A) THE DAY-ONE ANSWER — the recommendation we would give if the client forced us to answer today, in 3 sentences, plus a confidence label (gut / informed / evidence-backed).

B) LOAD-BEARING CLAIMS — the 4-6 claims that must all be true for that answer to hold. For each: why it matters, what we know today, and the single analysis that would prove it false.

C) KILL ORDER — the claims ranked by how cheaply they could be disproved and how badly the answer changes if they fail. Week one starts at the top of this list.

D) WHAT WOULD CHANGE OUR MIND — for each claim, the specific finding that would make us drop it, written before any data comes back so we cannot move the goalposts later.

E) ALTERNATIVE ANSWER — the most credible competing recommendation and the one claim that separates it from ours.

Inputs: [CLIENT ASK + DECISION THEY FACE] · [WHAT WE KNOW TODAY, WITH SOURCES] · [ENGAGEMENT LENGTH] · [CONSTRAINTS THE CLIENT HAS STATED]

Rules: Label every claim as known, assumed, or unknown, and do not invent client facts, market sizes, or benchmarks to make the answer look stronger. Verify each known claim against its source before the memo circulates to the team. Keep client-confidential material out of consumer AI tools beyond what I pasted, and anonymize the client in the memo.

Why it works — Answer-first is how strong teams avoid boiling the ocean, but it only works when the guess is written so it can lose. Listing the load-bearing claims, ranking them by how cheaply they can be killed, and writing the change-our-mind finding before data arrives turns a hunch into a test plan. The competing answer keeps the team honest about the one fact that actually decides between them.

Use AI for the ugly first draft, you for the judgment

The blank page is the tax. AI removes it. Your value moves to editing and judgment — the part that's actually yours.

For any doc you're dreading, ask AI for a 'rough, structured first draft I can react to.' Reacting to a draft is far easier than facing a blank page.

1. Give it the goal + your context block.
2. Ask for a rough structured draft.
3. Edit hard — keep 30%, rewrite the rest in your voice.